COMPOSITES

SCIENCE AND
TECHNOLOGY
Composites Science and Technology 65 (2005) 2300–2313
www.elsevier.com/locate/compscitech

Influence of different carbon nanotubes on the mechanical

properties of epoxy matrix composites – A comparative study
Florian H. Gojny 1, Malte H.G. Wichmann, Bodo Fiedler, Karl Schulte *

Technical University Hamburg-Harburg, Polymer Composites Section, Denickestrasse 15, 21073 Hamburg, Germany

Received 14 February 2005; received in revised form 14 April 2005; accepted 20 April 2005
Available online 16 June 2005

Abstract

Carbon nanotubes (CNTs) in general are considered to be highly potential fillers to improve the material properties of polymers.
However, questions concerning the appropriate type of CNTs, e.g., single-wall CNTs (SWCNT), double-wall CNTs (DWCNT) or
multi-wall CNTs (MWCNT), and the relevance of a surface functionalisation are still to be answered. This first part of the study
focuses on the evaluation of the different types of nanofillers applied, their influence on the mechanical properties of epoxy-based
nanocomposites and the relevance of surface functionalisation. The nanocomposites produced exhibited an enhanced strength and
stiffness and even more important, a significant increase in fracture toughness (43% at 0.5 wt% amino-functionalised DWCNT). The
influence of filler content, the varying dispersibility, the aspect ratio, the specific surface area and an amino-functionalisation on the
composite properties are discussed and correlated to the identified micro-mechanical mechanisms.

2005 Elsevier Ltd. All rights reserved.

Keywords: Carbon nanotubes; Epoxy matrix; Nanocomposites; Mechanical properties; Fracture toughness

1. Introduction to 1300 m2/g [9], as well as an aspect ratio in the range

In recent years, the development of nanocomposites
has become an attractive new subject in materials sci-
ence. Nanoparticles in general are regarded as high po-
tential fillers to improve the mechanical properties of
polymers. Furthermore, dependent on the applied type
of filler, nanoparticles can influence the electrical and
thermal conductivity of the final nanocomposite.
Interesting candidates with potentially unique prop-
erties are carbon nanotubes (CNTs) [1–4]. CNTs exhibit
an exceptionally high stiffness and strength [5–8], a
diameter dependent specific surface area (SSA) of up
*
Corresponding author. Fax: +49 40 42878 2002.
E-mail addresses: f.gojny@tuhh.de (F.H. Gojny), schulte@tuhh.de
(K. Schulte).
URL: http://www.tu-harburg.de/kvweb/ (K. Schulte).
1
Tel.: +49 40 42878 3238; fax: +49 40 42878 2002.
of several thousands. According to their graphitic struc-
ture, CNTs possess a high thermal conductivity and an
electrical conductivity, which can be either semi-con-
ducting or metal-like.
The combination of the previously mentioned mate-
rial properties makes CNTs highly desirable candidates
to improve the properties of polymers. Besides, the
development of CNT/polymer nanocomposites opens
new perspectives for multi-functional materials, e.g.,
conductive polymers with improved mechanical perfor-
mance and with a perspective of damage sensing and
‘‘life’’-monitoring.
In order to efficiently exploit the potential of CNTs to
improve the mechanical performance of polymers, one
has to be aware of aggravated challenges when compar-
ing with conventional micro-scaled filler particles. The
extraordinary large SSA of nanoparticles, being several
decades larger compared to conventional reinforcement
fibres (e.g., short carbon fibres: SSA  1 m2/g), leads to

0266-3538/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compscitech.2005.04.021
 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
special challenges, which can be summarised in (i) an
appropriate dispersion of the reinforcements in the ma-
trix, (ii) a sufficient interfacial bonding and (iii) receiving
representative information about nano-structural influ-
ences (CNT-structure-property relationship and nano-
(micro-) mechanical mechanisms).
The surface area of nanotubes can act as desirable
interface for stress transfer, but undesirably induces
strong attractive forces between the CNTs themselves,
leading to excessive agglomeration behaviour. The
SSA of CNTs is dependent on the diameter and the
number of sidewalls [9], where a maximum is provided
by SWCNTs. Additionally, SWCNTs have the largest
aspect ratio compared to CNTs consisting of multiple
layers. In order to minimise the SSA, SWCNTs form
aggregates of bonded and aligned CNT bundles, called
nano-ropes. These ropes, consisting of ten to hundreds
of individual tubes, are difficult to separate and infiltrate
with matrix.
The other extreme are MWCNTs. Having a much
larger diameter and consisting of several concentric
walls, these nanotubes provide a SSA of only 200 m2/g
or less. Therefore, MWCNTs exhibit a much better dis-
persibility, but they provide smaller interface for stress
transfer and a lower aspect ratio. Furthermore, the
stress transfer between the concentric layers has to occur
via interlayer shearing to be transferred by van der Wa-
als forces, which are relatively weak. As a conclusion,
MWCNTs can considered to be less efficient concerning
a mechanical reinforcement. We also have chosen to use
DWCNTs for this study, possessing a SSA of 600–
800 m2/g, giving a compromise between dispersibility
and reinforcing potential.
Various methods to disperse nanotubes in polymer
resins, such as stirring and sonication, have been re-
ported in the literature [10,11]. Most of these methods
are either limited in capacity or not powerful enough
to separate the agglomerates into individual nanotubes.
One common technique to distribute CNTs in epoxies is
the sonication technique. A pulsed ultrasound separates
the CNTs within agglomerates and disperses them in the
matrix effectively. However, this method is only man-
ageable for small batches due to the extreme reduction
of the vibrational energy with increasing distance from
the sonotrode. Another detrimental effect of this method
is the reported rupture of the CNTs, caused by the local
energy input [12–14], resulting in a reduction of the
effective tube length.
Retaining a high aspect ratio of the CNTs, enabling
an efficient load transfer to the nanotube, is of special
interest towards an improvement of the mechanical per-
formance of a CNT-reinforced polymer. The calander-
ing method, introduced previously [15], has turned out
to be effective with regard to the achieved dispersion
and exhibits the opportunity of up-scaling the capacity
to reach technical demands. Therefore, we believe this
2301
method to be a very promising approach to solve the is-
sue of dispersion of nanoparticles for resin systems in
the future.
The second issue in the development of CNT/poly-
mer composites is the interfacial adhesion between the
CNTs and the matrix polymer. A sufficient stress trans-
fer from the matrix to the tubes is required to efficiently
exploit the potential of CNTs as structural reinforce-
ment. The interfacial bonding between CNT and matrix
can be improved by functionalising the CNT-surface.
The introduction of tailored chemical groups (e.g.,
amino-, carboxyl- or glycidyl-groups for epoxies) enables
covalent bonding between CNTs and epoxy, improving
the interfacial stress transfer and positively affecting the
dispersibility of the nanofiller. The described effect of a
functionalisation on the mechanical properties has been
predicted by simulations [16] and also experimentally
proved by previous work [15,17,19].
The third main issue is a comprehensive knowledge
about the integrity of the CNTs resulting in improved
effectiveness on the mechanical and physical properties.
According to the production techniques, as chemical va-
pour deposition (CVD), electric arc-discharge method,
laser ablation and other catalytic processes, the nano-
tubes possess deviations of mechanical and physical
properties. Influencing parameters are the defect-density
and distribution (degree of graphitisation), the curva-
ture, the aspect ratio, the length- and diameter distribu-
tions, the density and the purity. Furthermore, one will
find numerous variations in CNT length and diameter
and a distribution of different chiralities in one batch
as well. As a conclusion, similarities can be drawn to
polymers, where it is a common and established proce-
dure to describe properties and structural features by
the use of distributions (chain length, molecular weight,
tacticity, etc.).
The function of CNTs as electrically conductive addi-
tive is generally understood and has already been com-
mercially exploited. The reinforcing potential of CNTs
on the other hand is still widely undeveloped and needs
further basic research. Progress reported in the literature
[15,17–29], can be summarised in that way, that there
are still difficulties in achieving a reinforcing effect of
CNTs in epoxy matrices, to prove their potential as
structural modifier.
Allaoui et al. [18] investigated the influence of
MWCNTs in a rubbery epoxy matrix. The addition of
up to 4 wt% MWCNTs led to a significant increase in
strength and YoungÕs modulus. However, the suggested
preparation method, using solvents (methanol), stirring
and evaporation, did not lead to an adequate distribu-
tion. Agglomerates in the dimension of several tens of
micrometers could still be detected by light microscopy.
In this context, the authors would like to mention, that
light microscopy may be a preliminary method in order
to get an idea of the overall distribution of agglomerates,
2302 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
but it is definitely not suitable to substantiate the discus-
sion about the dispersion of CNTs, due to the limited
resolution, insufficient for nanofillers.
The same group [19] reported an influence of length
and aggregate size of MWCNTs on the improvement
in mechanical and electrical properties of a standard
DGEBA-based nanocomposite. Composites containing
0.5–4 wt% CVD-grown MWCNTs exhibited an in-
creased YoungÕs modulus, but a reduction in fracture
strain. The reduction in fracture strain was explained
in terms of the existence of agglomerates, leading to lo-
cal defects enhancing early failure. Furthermore, a cer-
tain dependence of the mechanical behaviour on the
aggregate size could be found.
Liao et al. [20] investigated the influence of SWCNTs
on the thermo-mechanical properties of epoxy-based
nanocomposites. The composites were produced by the
use of sonication and the application of solvents and
surfactants to disperse the nanotubes. A dependence of
the achieved dispersion on the relative improvement of
the mechanical properties was observed. The prepara-
tion route was considered to be an effective method in
order to disperse CNTs. However, in their case, the
incorporation of SWCNTs led to a reduction of the
glass transition temperature, being explained by a reduc-
tion of the interfacial adhesion due to the use of a sur-
factant and the mentioned solvent.
Miygawa and Drzal [21] reported an increase in the
storage modulus of epoxy-based nanocomposites con-
taining fluorinated SWCNTs. A linearly decreasing glass
transition temperature with increasing filler content was
also observed. This result could indicate that a fluorina-
tion of CNTs may not be a suitable functionalisation to
significantly improve the interfacial adhesion. A correla-
tion of fluorinated SWCNTs to PTFE- or other matrix
incompatible particles could be drawn, which would in-
duce a similar behaviour.
The influence of oxo-fluorination of MWCNTs on
the fracture toughness of epoxy matrix nanocomposites
has been investigated by Park et al. [22]. A certain in-
crease in fracture toughness by SENB-testing (single
edge notch bending) was shown. The results were related
to the modified surface polarity of the CNTs, leading to
an improved bonding to the matrix by polar interac-
tions. In contrast to the neat fluorination, this function-
alisation generates additional hydroxyl groups on the
CNT surface, enabling hydrogen bonds to the epoxy
matrix.
Promising results, concerning a mechanical reinforce-
ment of epoxy resins by the use of CNTs were recently
reported by Zhu et al. [23,24]. They showed a significant
progress in improving the CNT dispersion and matrix
adhesion by adding 1 wt% alkylamino-functionalised
SWCNTs to an epoxy matrix thus, resulting in an in-
creased strength and YoungÕs modulus. The presented
functionalisation assimilates the polarity and directly
incorporates the SWCNTs into the epoxy network.
Their results were in agreement with our results
[15,17,25] using smaller quantities of amino-functional-
ised CNTs. In addition to the improvement in strength
and stiffness, an increased glass transition temperature
was observed. This substantiates previous results
[25,26], reporting a similar influence of the direct incor-
poration of CNTs into the matrix by covalent bonding.
Tailored functional groups induce the formation of
covalent bonds to the matrix material. We would like
to point out that the surface chemistry of carbon nano-
tubes (introduction of functional groups), could be re-
garded as the most important aspect in efficiently
exploiting the benefits of CNTs for a mechanical
reinforcement.
This brief description shows the diversity of possible
variables in developing CNT/epoxy nanocomposites.
A comparison of results reported is often difficult, due
to the usage of different epoxy matrices, with different
processing techniques and parameters, as well as the
choice of CNTs with different pre-treatments (e.g., func-
tionalisation), from various sources with again different
types and qualities. Presently, CVD is regarded as the
most relevant manufacturing process for CNTs, due to
the high production capacity fulfilling technical and in
the near future industrial demands as well [27,28].
In this part of the study, we describe the influence
of different types of CVD-grown CNTs (single-, dou-
ble- and multi-wall CNTs) on the mechanical proper-
ties of epoxy-based nanocomposites. We produced
nanocomposites using identical processing conditions,
by systematically varying CNT-type, filler content
and surface functionality. All results are compared
to reference materials, as the neat epoxy matrix and
a nanocomposite containing carbon black. Keeping
most variables constant, it was the aim of this study
to compare the influence of these different types of
CNTs on the mechanical properties and to appraise
the real potential of CNTs as structural modifiers of
epoxy-based composites. The modification of the elec-
trical and thermal conductivity by CNTs will be pre-
sented separately [32].
2. Experimental
2.1. Materials
2.1.1. Epoxy matrix system
The epoxy matrix used in this study consists of a
modified DGEBA-based epoxy resin (L135i) with an
amine hardener (H137i), supplied by Bakelite MGS Ku-
nstharzprodukte GmbH, Stuttgart/Germany. This
epoxy system is a standard resin for infusion processes
(e.g., RTM – resin transfer moulding) and it is charac-
terised by its low viscosity (gRT = 250 MPa s).
 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
2.1.2. Carbon nanotubes
All CNTs used in this comparative study were pro-
duced via the CVD-route:
(a) Single-wall carbon nanotubes
Single-wall carbon nanotubes were purchased from
Thomas Swan Ltd. & Co., UK. The SWCNTs exhibit
an average diameter of d < 2 nm and a length of some
micrometers.
(b) Double-wall carbon nanotubes (purified and amino-
functionalised)
The DWCNT (–NH2) were obtained from Nanocyl,
Namur/Belgium. They appear in an entangled cotton-
like form. These nanotubes consist of two graphitic
shells and have an average outer diameter of 2.8 nm
and a length of several micrometers. The amino-func-
tionalisation was accomplished by ball-milling purified
DWCNTs in ammonia. The degree of functionalisation,
specified by the manufacturer, is about one percent in
relation to the carbon content. It has to be mentioned
that the ball-milling process leads to a considerable
reduction of the effective length and unfortunately also
to a condensed packing of the CNTs.
(c) Multi-wall carbon nanotubes (purified and amino-
functionalised)
The MWCNTs were also obtained from Nanocyl.
For this study, we used thin MWCNTs (purified) and
amino-functionalised MWCNTs with an average outer
diameter of about 15 nm and a length of up to 50 lm.
The functionalisation process was similar to the one de-
scribed for DWCNTs.
2.1.3. Carbon black
We used nanocomposites containing highly conduc-
tive carbon black (CB) as reference material. The CB,
Printex XE2 from Degussa/Germany, is highly graphi-
tised and commonly used as conductive filler for anti-
static applications or as colour pigment for printing
ink. It is usually used to increase the electrical and ther-
mal conductivity of polymers, but also to enhance the
stiffness.
The CB tends to form aggregates, consisting of
bonded spherical primary particles with a diameter of
30 nm, forming again agglomerate structures. Carbon
black, being common and cheap nanoparticles, has
extensively been investigated in the past. We used CB
as reference in order to appraise the effect of the parti-
cle-shape on the achieved improvements concerning
the mechanical properties and, therefore, to obtain
information on the real potential of CNTs as mechani-
cal reinforcement for epoxy matrices.
2.2. Dispersion and processing parameters
The composites investigated in this study were all
manufactured in the same way. In order to disperse
the nanofillers in the epoxy resin, we first manually
2303
mixed the CNTs (CB) with the epoxy resin (without
hardener). This suspension was added batch-wise to a
mini-calander for final high shear mixing. The gap size
between the ceramic rolls of the mini-calander (Al2O3)
was 5 lm and the speed was set to 20 (1st roll), 60
(2nd roll) and 180 rpm (3rd roll). The dwell time of each
batch of suspension was 2 min. The suspension was
collected, mixed with the hardener for 10 min by intense
stirring, cured for 24 h at room temperature (RT) and
finally post-cured at 60 C for 24 h.
2.3. Mechanical characterisation
Tensile tests on nanocomposites were performed
according to DIN EN 527.1/2 with dog-bone specimens
using a Zwick universal tensile tester at a crosshead
speed of 1 mm/min. A statistical evaluation of the re-
sults was performed on the base of at least eight individ-
ually tested specimens of each sample.
The experimental investigation of the fracture tough-
ness was performed according to ASTM D 5045-99.
Compact tension (CT) specimens were tested at a cross-
head speed of 1.3 mm/min. The deformation of the sam-
ples was measured by detecting the crack opening
displacement (COD) using a highly sensitive extensome-
ter (MTS) with a gauge length of 5 mm. The sharp incip-
ient crack was introduced by carefully hammering a
sharp razor blade into the notch, whereas a crack initi-
ates in the zone of stress concentration in front of the
blade. A minimum of 6 specimens of each sample was
tested for statistical evaluation.
2.4. Electron microscopy
Scanning electron microscopy (SEM)-images were ta-
ken with a Leo FE-SEM 1530 at different acceleration
voltages (0.9–1.5 kV). The samples were investigated
without gold-sputtering to avoid covering of the nano-
scaled reinforcements. The qualitative determination of
the dispersion was accomplished by transmission elec-
tron microscopy (TEM), using a Philips EM 400 at
120 kV, on thin films (50 nm) of nanocomposites.
The films were obtained by ultra-microtome cutting at
room temperature.
3. Results and discussion
3.1. Influence of SSA and surface functionalisation of the
CNTs on the dispersibility in epoxy resin
Dispersing nanoparticles in a chosen matrix system is
a challenging process. As already mentioned in the intro-
duction, nanoparticles exhibit a SSA, which is several
orders of magnitude larger than that of conventional
fibre-reinforcements (e.g., SWCNT: SSA  1300 m2/g).
2304 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
In order to homogeneously disperse nanoparticles in the
epoxy and to properly impregnate the potential surface
area of the nanofiller, a huge energy input has to be en-
sured to overcome the nanotube–nanotube adhesion by
van der Waals-forces. A local introduction of the re-
quired energy, as performed by sonication, will lead to
a damage of the nanotube structure and to a reduction
of the effective length [12–14]. We could prove that the
application of a mini-calander is an effective method to
disperse CNTs and CB in epoxy matrix systems [17],
thus, all nanocomposites investigated in this study were
manufactured using this method. Presently, TEM is the
only method to attain qualitative information concern-
ing a distribution of nanoparticles in a polymer matrix.
Therefore, we investigated the produced composites by
TEM. In Fig. 1(a)–(f), representative TEM micrographs
of the achieved dispersion for the investigated CNTs are
shown.
Fig. 1. TEM-images of the manufactured composites. Achieved state
of dispersion for: (a) epoxy/MWCNT; (b) epoxy/MWCNT–NH2;
(c) epoxy/DWCNT; (d) epoxy/DWCNT–NH2; (e) epoxy/SWCNT;
(f) epoxy/CB.
As it can be seen from Fig. 1(a) and (b), both
MWCNT and MWCNT–NH2, respectively, exhibit a
high dispersibility in the epoxy matrix. The homoge-
neous dispersion in the matrix was achieved by calan-
dering with only very few agglomerates remaining. The
MWCNTs exhibit the smallest SSA (200 m2/g) of the
investigated tubes, tantamount to the lowest attractive
forces and agglomeration behaviour.
The amino-groups seem to stabilise the CNT-disper-
sion by stronger interactions with the epoxy matrix. This
can be attributed to the increased polarity of the CNTs
by the functional groups and the possible reaction of the
amino-groups with the epoxy resin, forming covalent
bonds. Fig. 1(b) indicates a very good matrix adhesion
of the amino-functionalised MWCNTs. Some of the im-
aged CNTs, stick out of the plane of the prepared films.
These CNTs have not been cut by the diamond blade,
but pulled out. However, the remaining epoxy layer
coating the nanotubes indicates a very strong matrix
adhesion.
The described effects were also observed for the
DWCNTs. According to the larger surface area of
700 m2/g, the interactive forces between the DWCNTs
are significantly larger, thus, requiring more energy to
disperse the aggregated nanotubes in the epoxy matrix.
The achieved dispersions are shown in Fig. 1(c) and
(d). The two different types of DWCNTs, non-function-
alised DWCNTs (Fig. 1(c)) and amino-functionalised
DWCNTs (Fig. 1(d)) show a mixed structure of bro-
ken-up agglomerates and isolated nanotubes. The align-
ment of the DWCNTs to be seen in Fig. 1(c) and (d)
could be caused either by ultra-microtome cutting, as re-
ported by Ayajan et al. [29] for similar epoxy/CNT-
nanocomposites, or by the induced shearing.
The achieved dispersion for SWCNTs in the epoxy
matrix is shown in Fig. 1(e). The TEM-micrograph
shows a mixture of separated SWCNTs, as well as
mostly small agglomerates with a diameter of 50–
300 nm. Furthermore, few larger agglomerates exhibit-
ing a diameter of about two micrometer were observed.
A further improved dispersion should be possible for
all CNTs using this calandering method, but a detailed
parameter study is required in this case. So far the ques-
tion concerning the favourable state of dispersion, it is
still unsolved. Do we have to achieve a structure of com-
pletely separated CNTs homogeneously dispersed in the
epoxy matrix or not? Fact is, that a good impregnation
of the nanofillers is the prerequisite for an efficient
exploitation of the potential benefit from CNTs as struc-
tural reinforcement. A combination of impregnated
agglomerates and isolated nanotubes can be considered
to be a favourable state of dispersion for at least some
purposes. A detailed discussion on this topic is accom-
plished in Section 3.3.
Fig. 1(f) is a TEM-micrograph of the reference mate-
rial containing carbon black. As it can be seen in the
 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
Fig. 1 (continued)
micrograph, the CB was separated into the aggregate
structure. These aggregates are homogeneously dis-
persed in the matrix system. The generation of high
shear forces during the calandering process reflects itself
in the separation of primary particles from the aggre-
gates, frequently observed by TEM.
The achieved dispersion reveals the applicability of a
calandering technique to efficiently disperse various
nanoparticles in epoxies and probably also in other li-
quid polymers and polymer solutions.
3.2. Evaluation of the mechanical properties of CNT/
epoxy composites
The reinforcement potential of the CNTs can only be
activated if an effective load transfer from the surround-
ing epoxy matrix into the nanotube and reverse is possi-
ble. Therefore, a good impregnation with the matrix,
together with a strong interfacial adhesion has to be en-
sured. According to the fact that a stress transfer can
only occur via the outermost layer of a nanotube, one
2305
can conclude a maximum potential of SWCNTs con-
cerning mechanical reinforcement. However, the larger
the provided interface, the more serious the dispersion
conditions, thus, aggravating the impregnation. Alterna-
tives could be CNTs exhibiting a smaller SSA, which led
to a better dispersibility.
In this section, we compare the results of the mechan-
ical testing and discuss parameters influencing the
mechanical reinforcement by CNTs. The results of the
neat epoxy and the analogous composites containing
the same filler contents CB as those of the CNTs are
shown as reference in all of the following figures. All
mechanical properties are summarised in Table 1.
Fig. 2 shows the ultimate tensile strength (UTS) of
the nanocomposites containing non-functionalised
(Fig. 2(a)) and functionalised CNTs (Fig. 2(b)) as func-
tion of filler content, respectively. The UTS of the neat
epoxy matrix is shown in the form of the dashed line
for comparison.
In general one can say, that the addition of small
amounts SWCNTs and DWCNTs increases the UTS
2306 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313

Table 1
Mechanical properties of nanoparticle reinforced composites
Filler type/content (wt%) YoungÕs modulus (MPa) Ultimate tensile strength (MPa) Fracture toughness KIc
(MPa m1/2)
Epoxy 0.0 2599 (±81) 63.80 (±1.09) 0.65 (±0.062)
Epoxy/CB 0.1 2752 (±144) 63.28 (±0.85) 0.76 (±0.030)
0.3 2796 (±34) 63.13 (±0.59) 0.86 (±0.063)
0.5 2830 (±60) 65.34 (±0.82) 0.85 (±0.034)
Epoxy/SWCNT 0.05 2681 (±80) 65.84 (±0.64) 0.72 (±0.014)
0.1 2691 (±31) 66.34 (±1.11) 0.80 (±0.041)
0.3 2812 (±90) 67.28 (±0.63) 0.73 (±0.028)
Epoxy/DWCNT 0.1 2785 (±23) 62.43 (±1.08) 0.76 (±0.043)
0.3 2885 (±88) 67.77 (±0.40) 0.85 (±0.031)
0.5 2790 (±29) 67.66 (±0.50) 0.85 (±0.064)
Epoxy/DWCNT–NH2 0.1 2610 (±104) 63.62 (±0.68) 0.77 (±0.024)
0.3 2944 (±50) 67.02 (±0.19) 0.92 (±0.017)
0.5 2978 (±24) 69.13 (±0.61) 0.93 (±0.030)
Epoxy/MWCNT 0.1 2780 (±40) 62.97 (±0.25) 0.79 (±0.048)
0.3 2765 (±53) 63.17 (±0.13) 0.80 (±0.028)
0.5 2609 (±13)a 61.52 (0.19)a a

Epoxy/MWCNT–NH2 0.1 2884 (±32) 64.67 (±0.13) 0.81 (±0.029)

0.3 2819 (±45) 63.64 (0.21) 0.85 (±0.013)
0.5 2820 (±15) 64.27 (±0.32) 0.84 (±0.028)
a
High viscosity disabled degassing – composite contained numerous voids.

moderately, whereas MWCNTs lead to a slight reduc-
tion. CB, as reference material, does not significantly
influence the UTS of the epoxy matrix at very low filler
contents.
The relatively large improvement in UTS by
SWCNTs compared to the other nanofillers can be ex-
plained by the high specific mechanical properties and
the highest aspect ratio of the SWCNTs. Combined with
the high SSA, SWCNT do not have inactive internal lay-
ers, they exhibit the highest potential to improve the
strength of materials. DWCNTs consisting of two con-
centric layers, exhibit a smaller SSA and lower aspect ra-
tio, but do not agglomerate as pronounced as SWCNTs.
The composite containing 0.3 wt% DWCNT exhibits a
maximum UTS (67,8 MPa), which did not further in-
crease with higher tube loadings.
Neither the addition of small amounts of MWCNT,
nor MWCNT–NH2, do significantly affect the UTS of
the epoxy resin. An explanation for the observed behav-
iour can be found by considering the concentric struc-
ture of MWCNTs. Due to the absence of stress
transfer to internal layers, only the outermost layers of
MWCNTs contribute to the mechanical reinforcement.
Based on geometrical calculations, it is possible to show
for an imaginary nanotube with an outer diameter of
15 nm, an inner diameter of 4 nm and a layer distance
of 0.32 nm, an effective surface area of only 9% of the
total weight contributes to the reinforcement of the
epoxy matrix. The slight reduction of the UTS by
MWCNTs compared to the neat epoxy matrix and the
relatively lower increase for SWCNTs and DWCNTs
at higher filler contents can be attributed to an increas-
ing amount of improper impregnated agglomerates, act-
ing as imperfections in the composite, inducing early
failure.
The UTS of composites containing amino-functional-
ised CNTs is summarised in Fig. 2(b). The amino-func-
tionalisation of the CNTs improves the dispersibility by
increasing the surface polarity and induces an incorpo-
ration of the nanotubes into the matrix network, thus,
resulting in a generally larger improvement of the
UTS. A content of 0.5 wt% DWCNT–NH2 led to an in-
crease in UTS from 63.7 MPa (epoxy) to 69.2 MPa
(+10%). The UTS of the composites containing
MWCNT–NH2 remains on the level of the neat epoxy.
The more efficient improvement of the UTS by function-
alised CNTs proves the importance of a chemical sur-
face functionalisation to integrate the CNTs by
covalent bonding in the network structure of the epoxy
matrix. This is an effective method to significantly im-
prove the interfacial adhesion, making functionalisation
to a key issue in the development of CNT-reinforced
epoxy composites.
Fig. 3 illustrates the corresponding YoungÕs moduli
of composites containing non-functionalised (Fig. 3(a))
and amino-functionalised CNTs (Fig. 3(b)).
The addition of any kind of nanofiller increases the
stiffness of the matrix material, as long as a proper
impregnation with the matrix and a sufficient adhesion
is ensured. The neat epoxy matrix exhibits a YoungÕs
modulus of 2599 MPa. The present results
for SWCNT/epoxy composites show an almost linear
 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
Fig. 2. Ultimate tensile strength of epoxy-based composites containing
(a) non-functionalised and (b) amino-functionalised nanoparticles.
increase of the YoungÕs modulus in relation to the filler
content, yielding in a maximum of 2812 MPa at 0.3 wt%
under the given processing conditions.
The YoungÕs modulus of MWCNT/epoxy composites
increases at contents of 0.1 wt% MWCNT and de-
creased later for higher MWCNT loadings. A possible
explanation can be given for this behaviour by assuming
a similar state of high quality dispersion for all nano-
composites after the calandering process. An increasing
amount of agglomerates in the cured composites was
observed for the CNT/epoxy composites above filler
contents of / > 0.1 wt%. We propose these agglomer-
ates to be a result of reagglomeration. According to
the smaller number of particles in MWCNT/epoxy com-
posites, compared to SWCNTs, a reagglomeration of
MWCNTs does reduce the YoungÕs modulus more
significantly.
DWCNT showed stronger improvements in the
mechanical properties, yielding in a maximum YoungÕs
2307
Fig. 3. YoungÕs modulus of epoxy-based composites containing (a)
non-functionalised and (b) amino-functionalised nanoparticles.
modulus of 2885 MPa (+11%) at 0.3 wt% DWCNTs,
but decreasing again for higher filler contents. This effect
can again be attributed to the reagglomeration, but due
to the larger number of particles we observe a shift to
higher filler contents.
The amino-functionalised nanotubes turned out to be
more efficient concerning an improvement of the
YoungÕs modulus, thus again proving the relevance of
surface functionalisation. Amino groups located onto
the CNT surface will react under formation of covalent
bonds with the epoxy, thus, resulting in a significantly
enhanced interfacial adhesion. We observed a maximum
increase of the YoungÕs modulus for a composite con-
taining 0.5 wt% DWCNT–NH2 (2978 MPa) by +15%
in relation to the neat epoxy resin (2599 MPa). How-
ever, the less efficient improvement at higher filler con-
tents can be referred to an increasing amount of
observed agglomerates and demonstrates the necessity
of a proper impregnation with the matrix.
2308 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
A similar behaviour, when compared to non-func-
tionalised MWCNTs, was observed for MWCNT–
NH2. These composites showed a stronger enhancement
of the YoungÕs modulus, but stagnating on the achieved
level at higher filler contents. The decreasing efficiency
can again be related to the reagglomeration, which
was less intensive as for the non-functionalised
MWCNTs.
An estimation of the relative improvement of
strength and stiffness by any kind of filler should be per-
formed according to its volume fraction. However, the
density of CNTs is dependent on the CNT-diameter
and, therefore, required for a determination of the vol-
ume fraction. Thostensen and Chou [5] suggested an ap-
proach which considers the inner- and outer diameter of
CNTs and its distributions, as well as possible defects in
the CNT structure. Unfortunately, reliable and statisti-
cally validated data (length, diameters, etc.) are difficult
to obtain due to the nanometric dimensions.
The effect of a surface functionalisation becomes vis-
ible by directly comparing the stress–strain (r/e-) curves
of the tested samples. Fig. 4 shows representative r/e-
curves of the epoxy resin and two composites containing
each 0.3 wt% of DWCNTs and DWCNT–NH2,
respectively.
The presence of 0.3 wt% DWCNTs leads to an in-
crease in UTS and strain to failure, when compared to
the neat epoxy resin. However, using the same
DWCNTs, but with an amino-functionalised surface,
an even larger improvement can be observed. Regarding
the area under the r/e-curves as the fracture work of the
composites, first evidences were given for improved frac-
ture toughness. A detailed discussion is accomplished in
the following section.
Fig. 4. Comparison of the stress/strain-behaviour of the neat epoxy
and nanocomposites containing 0.3 wt% DWCNTs and 0.3 wt%
DWCNT–NH2.
In conclusion, we assume the provided surface area,
in combination with the specific mechanical properties
of the nanofiller and the aspect ratio to be the determi-
nant factors for the mechanical reinforcement of epoxy
matrices by nanoparticles. However, a complete impreg-
nation of the surface area has to be ensured, in order to
enable an efficient load transfer between matrix and
reinforcement. A chemical functionalisation of the
CNT surface improves the dispersibility and the interfa-
cial adhesion and can, therefore, be a useful tool in
order to realise this impregnation. Presently, DWCNT–
NH2 reinforced epoxy shows the largest improvement of
strength and stiffness under the given processing
conditions.
An even better improvement should be possible using
functionalised SWCNTs. Realising an appropriate dis-
persion and matrix adhesion by integration into the net-
work structure, functionalised SWCNT provide a
maximum surface and, therefore, exhibit the largest po-
tential to improve the mechanical properties of
polymers.
3.3. Potential of CNTs to improve the fracture toughness
of an epoxy matrix
Epoxy resins are favourable and commonly used ma-
trix systems for a huge variety of applications. They pro-
vide a high strength and stiffness, good thermal stability,
high thermo-mechanical stability and good resistivity
against chemicals, but a relatively low toughness. The
fracture toughness of polymers, tantamount to an im-
proved resistivity against crack propagation, thus a
higher damage tolerance, is a crucial and limiting factor
for the design of structural components, especially with
regard to the long-term behaviour (fatigue) [30]. There-
fore, an improvement of the fracture toughness of epox-
ies is a desirable goal of research activities.
A comprehensive knowledge about the influence of
nanoparticles on the micromechanics is required in or-
der to explain the observed toughening effect of nano-
particles. Recently, we published an overview over the
influence of nanoparticles on the fracture toughness
and the related micro-mechanical mechanisms [31]. In
summary, the most important mechanisms leading to
an increase in fracture toughness are: (i) localised inelas-
tic matrix deformation and void nucleation; (ii) particle/
fibre debonding; (iii) crack deflection; (iv) crack pinning;
(v) fibre pull-out; (vi) crack tip blunting (or crack tip
deformation); (vii) particle/fibre deformation or their
rupture at the crack tip. Additionally, the overall size
of the plastic deformation zone has to be considered.
The fracture surfaces of composites give first infor-
mation about fracture mechanisms and the influence
of particle modification on the fracture behaviour. Fig.
5(a) and (b) shows SEM images of fracture surfaces of
the neat epoxy resin (Fig. 5(a)) and a representative
 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
Fig. 5. SEM-micrographs of a fracture surfaces at low magnifications
(1000·), showing (a) the epoxy and (b) a DWCNT–NH2/epoxy
composite. The composite containing CNTs exhibits a significantly
rougher fracture surface compared to the neat epoxy, indicating a
toughening effect of the nanoparticles.
nanocomposite containing 0.1 wt% DWCNT–NH2 (Fig.
5(b)). The imaged fracture surfaces were taken from
similar areas of the failed specimens. The general tough-
ening effect of (nano-) particles reflects itself in a signif-
icantly larger roughness of the fracture surface of the
sample containing CNTs. A similar effect has been ob-
served for all nanofillers, indicating this effect being re-
lated to the general existence of particle reinforcements.
Nanoparticles in general, and CNTs in special, pos-
sess a certain potential to improve the fracture tough-
ness of epoxy systems due to their mechanical
properties, the fibre-shaped structure (high aspect ratio)
and the huge interface provided [13]. In this section, we
present the results of fracture-toughness-tests performed
on CT-specimens. The results are displayed in Fig. 6,
and will be discussed in correlation with the identified
micro-mechanical mechanisms.
The neat epoxy resin has a fracture toughness of
KIc = 0.65 MPa m1/2. Non-functionalised nanoparticles
generally increase the fracture toughness of the epoxy
2309
Fig. 6. Experimentally obtained fracture toughness of epoxy-based
composites containing (a) non-functionalised and (b) amino-function-
alised nanoparticles.
matrix significantly at very low filler contents, as shown
in Fig. 6(a). The relative improvement of the KIc-value is
not dependent on the particle-shape and, therefore, the
main fracture mechanical mechanism leading to the en-
hanced fracture toughness could be related to the huge
surface area of the nanoparticles. As discussed in Sec-
tion 3.1, we observed a partly agglomerated dispersion
for all nanofillers, which leads to the conclusion that
the localised inelastic matrix deformation, void nucle-
ation and crack deflection at the agglomerates are the
dominating toughening mechanisms.
A certain dependence of the provided surface area of
the nanofillers on the toughening capacity could be
found. In tendency, large surface areas lead to a more
efficient improvement of fracture toughness. The de-
crease in fracture toughness (e.g., SWCNT at
0.3 wt%), observed at higher filler contents, is related
to the reagglomeration. For all mechanical characteris-
tics, an exploitation of the theoretical surface area of
2310 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
the nanofiller as interface to the epoxy matrix is related
to the dispersion and matrix impregnation. Thus, the
interface is playing a major role concerning toughening
of materials. According to the mentioned toughening
mechanisms, we suggest a combination of individually
separated dispersed nanotubes and well-impregnated
small (sub-micrometer) agglomerates to be a favourable
state of dispersion.
The amino-functionalisation of CNTs improves their
dispersibility (see Section 3.1) and increases the interfa-
cial adhesion, leading to a higher effective interfacial
area in the composite. However, a too strong interfacial
adhesion between matrix and reinforcement negatively
affects the toughness by suppressing interfacial failure,
which is prerequisite for many toughening mechanisms.
Our experimental results, displayed in Fig. 6(b),
prove a positive influence of an amino-functionalisation
of the CNT surface on the fracture toughness. The mea-
sured fracture toughness of the composite containing
0.3 wt% DWCNT–NH2 (KIc = 0.92 MPa m1/2) was
42% higher than the neat epoxy (KIc = 0.65 MPa m1/2).
Fig. 7. SEM-micrographs of DWCNT(–NH2)/epoxy composites. A
crack, induced by etching, is bridged by (a) amino-functionalised and
(b) non-functionalised double-wall carbon nanotubes. Crack-bridging
is one micro-mechanical mechanism leading to the observed increase in
fracture toughness. The bridged gap length is about 500–1000 times the
average diameter of the nanotubes.
Besides the improved dispersibility and impregnation
of the surface, we also relate this effect to the integration
of CNTs into the epoxy-network structure. The rela-
tion of amino-groups to the number of surface carbon
atoms of the CNTs is about 1/100, according to the
information of the manufacturer. Furthermore, it was
shown that these functional groups appear in localised
domains onto the surface, thus, locally bonding the
CNT to the matrix. The localisation of functional groups,
especially at the ends of the CNTs, is a result of the func-
tionalisation process. A ball milling, as used in our case,
will break the CNTs and generate highly reactive ends,
which immediately react with the surrounding solvent
(amines). The situation described allows certain mobility
and reorientation in the matrix under stress situations,
which consumes more energy in case of interfacial fail-
ure, due to the additional rupture of the covalent bonds
to the epoxy-molecules. The high strain to failure of
CNTs [6,7] in relation to the epoxy enables a huge defor-
mation of the nanotube, which is several times that of
the epoxy matrix. This also allows a crack bridging of
Fig. 8. SEM-micrographs showing small agglomerates of (a) carbon
black/epoxy and (b) DWCNT–NH2/epoxy composites. Shear-bands
around the agglomerates indicate a major contribution of plastic
matrix deformation and void nucleation to the fracture toughness of
nanoparticle filled epoxy. This mechanism is independent from the
particle shape.
 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313 2311

Fig. 9. Schematic description of possible fracture mechanisms of CNTs. (a) Initial state of the CNT; (b) pull-out caused by CNT/matrix debonding
in case of weak interfacial adhesion; (c) rupture of CNT – strong interfacial adhesion in combination with extensive and fast local deformation; (d)
telescopic pull-out – fracture of the outer layer due to strong interfacial bonding and pull-out of the inner tube; (e) bridging and partial debonding of
the interface – local bonding to the matrix enables crack bridging and interfacial failure in the non-bonded regions.

functionalised DWCNTs (Fig. 7(a)), as well as non-
functionalised DWCNTs in the epoxy (Fig. 7(b)), as ob-
served by SEM.
The imaged cracks were induced into the composite
by stresses due to the drying of the composite after sur-
face etching with sulphuric acid. The imaged surface
cracks were not directly formed during etching, due to
the short etching time of only 60 s. We could instead ob-
serve the generation of the surface cracks inside the
SEM as a result of vacuum drying and thermal stresses.
This method allows the investigation of the composite
by SEM underneath the surface.
A comparison to the CB-containing epoxy compos-
ites underlines the advantage of fibrous over spherical
particles. Toughening mechanisms like localised inelas-
tic matrix deformation and void nucleation, interfacial
debonding and crack deflection at agglomerates are
accessible for all particles, regardless the particle shape.
These mechanisms could be identified by SEM investiga-
tions on the nanocomposites (Fig. 8(a) and (b)). We ob-
served shear-bands around the agglomerates of CB (Fig.
8(a)) and DWCNT–NH2 (Fig. 8(b)), thus, indicating
void nucleation caused by plastic deformation of the
matrix to be a major toughening mechanism.
In case of a nano-scaled reinforcement, we observe
toughening mechanisms occurring at two different
dimensional levels. A toughening of the epoxy by crack
deflection at agglomerates, crack pinning, crack blunt-
ing and the extension of the plastic deformation zone
in general, are (under consideration of the dimensions)
micro-mechanical mechanisms. Other toughening mech-
anisms like inelastic matrix deformation and void
nucleation, interfacial debonding, pull-out and crack-
bridging do exist in macroscopic and microscopic scale,
but according to the nano-sized structure of CNTs also
in the nano-scale and additionally further mechanisms
are possible, illustrated in Fig. 9.
A reliable quantitative description is presently diffi-
cult. It is yet not clear whether mathematical approxi-
mations based on macro-/microscopic approaches are
also valid to explain nano-mechanical processes or
not. Therefore, Fig. 9 schematically describes possible
processes and mechanisms, explaining the deformation
and failure of CNTs bridging a crack.
The initial situation of the CNT in an ideal case
(Fig. 9(a)) is a completely impregnated and isolated
embedded in the matrix. In case of a crack the mech-
anisms Fig. 9(b)–(e) can occur, depending on the
interfacial adhesion and the mechanical properties of
the CNTs. In case of a weak interfacial bonding sim-
ple pull-out of the CNT from the matrix occurs (Fig.
9(b)). In contrast, a very strong bonding between
CNT and matrix will lead to a complete rupture of
the CNT (Fig. 9(c)) or to fracture of the outer layer
and a telescopic pull-out of the inner tube(s) (Fig.
9(d)). A spatial bonding of the reactive groups at
the interface, as suggested in our case, enables a par-
tial debonding of the interface, but allows a crack
bridging (Fig. 9(e)). This bridging suppresses a further
crack opening. However, increasing stresses would fi-
nally lead to a failure of the CNT, according to Fig.
9(c) and (d).
4. Summary and conclusions
Nanocomposites consisting of a DGEBA-based
epoxy matrix system and different types of carbon nano-
tubes were produced by calandering. The composites
were investigated regarding the dispersibility of the
CNTs in the matrix and the efficiency concerning a
mechanical reinforcement. The obtained data were dis-
cussed in terms of influencing parameters as (i) specific
surface area (SSA), (ii) chemical functionalisation of
2312 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
the surface, and (iii) particle shape and correlated to
identified micro-mechanical mechanisms.
CNTs in general exhibit a certain potential to im-
prove the mechanical properties of the tested epoxy ma-
trix systems. A correlation between the provided surface
area and the dispersibility was found. The dispersibility
turned out to be more difficult, the larger the SSA of the
CNTs. A functionalisation with amino-groups improved
the dispersibility in all cases and stabilised the CNT-sus-
pension, thus, reducing the reagglomeration. This effect
was attributed to an assimilated polarity of the surface
and to a possible reaction between the introduced ami-
no-groups and the epoxy resin, thus, integrating the
CNTs into the epoxy-network structure.
The potential of CNTs to improve strength and
stiffness can be correlated to the specific mechanical
properties, the aspect ratio and the SSA. Thus, the
exploitation of the provided interface is playing a major
role concerning the improvement of mechanical proper-
ties. The dispersion of the CNTs and the interfacial
adhesion to the epoxy matrix are key issues in the
development of nanocomposites. A chemical function-
alisation of the CNT surface is certainly an important
approach to improve both interfacial bonding and
dispersibility. Small amounts of carbon nanotubes in
epoxies lead to increased mechanical properties. The
most significant improvements of strength (+10%), stiff-
ness (+15%) and especially fracture toughness (+43%)
were attained with amino-functionalised DWCNTs at
0.5 wt% filler content under the given processing
conditions.
Micro-mechanical mechanisms affecting the fracture
toughness (e.g., plastic deformation of the epoxy matrix
and nucleation of voids, crack deflection at agglomer-
ates, etc.) were identified. These micro-mechanical
toughening mechanisms observed were regardless the
particle-shape. In addition to this, nanomechanical
mechanisms, related to the fibre-shape of CNTs contrib-
ute to an enhancement of the fracture toughness. All
nanotube/epoxy composites, particularly the amino-
functionalised DWCNTs, being integrated in the epoxy
network, show a significant contribution of the bridging
mechanism to the fracture toughness. We suggest mech-
anisms explaining the increased fracture toughness of
epoxy-based composites, despite a strong matrix bond-
ing of the amino-functionalised CNTs. Commonly a
very strong interfacial bonding is a adverse to an
enhancement of the fracture toughness by using macro-
and microscopic particles.
In conclusion, carbon nanotubes enable the develop-
ment of a new generation of materials with multi-func-
tional properties, such as a combination interesting
physical properties [32] together with improved mechan-
ical performance. CNTs are a valuable chemical additive
for the modification of epoxies and other polymers.
Especially, the combination with conventional fibre-
reinforcements can be a promising approach for future
perspectives in composite applications [33–35].
Acknowledgements
The companies Bakelite MGS Kunstharzprodukte
GmbH/Stuttgart and Exakt Vertriebs GmbH/Hamburg,
Germany are acknowledged for the supply of the epoxy
resin and the mini-calander, respectively. The European
Commission (Scientific-Network: ‘‘Carbon Nanotubes
for Future Industrial Composites: theoretical potential
versus immediate application (CNT-Net)’’; Contract
No. G5RT-CT-2001-050206) is gratefully acknowledged
for financial support.
References
[1] Iijima S. Helical microtubules of graphitic carbon. Nature
1991;354:56–8.
[2] Oberlin A, Endo M. Filamentous growth of carbon through
benzene decomposition. J Cryst Growth 1976;32:335–49.
[3] Nesterenko AM, Kolesnik NF, Akhmatov YS, Sukhomlin VI,
Prilutski OV. Metals 3, UDK 869.173.23, News of the academy of
science, USSR; 1982. p. 12–6.
[4] Thostenson ET, Ren Z, Chou TW. Advances in the science and
technology of carbon nanotube and their composites: a review.
Compos Sci Technol 2001;61:1899–912.
[5] Thostenson ET, Chou TW. On the elastic properties of carbon
nanotube-based composites: modeling and characterization. J
Phys D: Appl Phys 2003;36:573–82.
[6] Yu MF, Lourie O, Dyer MJ, Moloni K, Kelly TF, Ruoff RS.
Strength and breaking mechanism of multiwalled carbon nano-
tubes under tensile load. Science 2000;287:637–40.
[7] Yu MF, Files BS, Arepalli S, Ruoff RS. Tensile loading of ropes
of single wall carbon nanotubes and their mechanical properties.
Phys Rev Lett 2000;84:5552–5.
[8] Li C, Chou TW. Elastic moduli of multi-walled carbon nanotubes
and the effect of van der Waals forces. Compos Sci Technol
2003;63:1517–24.
[9] Peigney A, Laurent Ch, Flahaut E, Bacsa RR, Rousset A. Specific
surface area of carbon nanotubes and bundles of carbon
nanotubes. Carbon 2001;39:507–14.
[10] Sandler J, Shaffer MSP, Prasse T, Bauhofer W, Schulte K, Windle
AH. Development of a dispersion process for carbon nanotubes in
an epoxy matrix and the resulting electrical properties. Polymer
1999;40:5967–71.
[11] Schulte K, Gojny FH, Fiedler B, Broza G, Sandler J. Carbon
nano tube reinforced polymers. A state of the art – review. In:
Friedrich K, Fakirov S, Zhang Z, editors. Polymer-composites –
from nano to macro scale. Dordrecht: Kluwer Publishers; 2005.
p. 3–23.
[12] Lu KL, Lago M, Chen YK, Green MLH, Harris PJF, Tsang SC.
Mechanical damage of carbon nanotubes by ultrasound. Carbon
1996;34:814–6.
[13] Shelimov KB, Esenaliev RO, Rinzler AG, Huffman CB,
Smalley RE. Purification of single-wall carbon nanotubes by
ultrasonically assisted filtration. Chem Phys Lett 1998;282:
429–34.
[14] Zhang M, Yudasaka M, Koshio A, Iijima S. Effect of polymer
and solvent on purification and cutting of single-wall carbon
nanotubes. Chem Phys Lett 2001;349:25–30.
 F.H. Gojny et al. / Composites Science and Technology 65 (2005) 2300–2313
[15] Gojny FH, Wichmann MHG, Köpke U, Fiedler B, Schulte K.
Carbon nanotube-reinforced epoxy-composites – enhanced stiff-
ness and fracture toughness at low nanotube contents. Compos
Sci Technol 2004;64:2363–71.
[16] Frankland SJV, Caglar A, Brenner DW, Griebel M. Molecular
simulation of the influence of chemical cross-links on the shear
strength of carbon nanotube–polymer interfaces. J Phys Chem B
2002;106:3046–8.
[17] Gojny FH, Nastalczyk J, Roslaniec Z, Schulte K. Surface
modified carbon nanotubes in CNT/epoxy-composites. Chem
Phys Lett 2003;370:820–4.
[18] Allaoui A, Bai S, Cheng HM, Bai JB. Mechanical and electrical
properties of MWNT/epoxy composite. Compos Sci Technol
2003;62:1993–8.
[19] Bai JB, Allaoui A. Effect of the length and the aggregate size of
MWNTs on the improvement efficiency of the mechanical and
electrical properties of nanocomposites – experimental investiga-
tion. Composites A 2003;34:689–94.
[20] Liao YH, Marietta-Tondin O, Liang Z, Zhang C, Wang B.
Investigation of the dispersion process Of SWNTs/SC-15 epoxy
resin nanocomposites. Mater Sci Eng A 2004;385:175–81.
[21] Miyagawa H, Drzal LT. Thermo-physical and impact properties
of epoxy nanocomposites reinforced by single-wall carbon nano-
tubes. Polymer 2004;45:5163–70.
[22] Park SJ, Joeng HJ, Nah C. A study of oxofluorination of
multi-walled carbon nanotubes on mechanical interfacial
properties of epoxy matrix nanocomposites. Mater Sci Eng
A 2004;385:13–6.
[23] Zhu J, Kim JD, Peng H, Margrave JL, Khabashesku VN, Barrera
EV. Improving the dispersion and integration od single-walled
carbon nanotubes in epoxy composites through functionalization.
Nano Lett 2003;3:1107–13.
[24] Zhu J, Peng H, Rodriguez-Macias F, Margrave JL, Khabashesku
VN, Imam AM, et al. Reinforcing epoxy composites through
2313
covalent integration of functionalized nanotubes. Adv Funct
Mater 2004;14:643–8.
[25] Gojny FH, Schulte K. Functionalisation effect on the thermo
mechanical behaviour of multi-wall carbon nanotube/epoxy-
composites. Compos Sci Technol 2004;64:2303–8.
[26] Gong X, Liu J, Baskaran S, Voise RD, Young JS. Surfactant
assisted processing of carbon nanotube/polymer composite. Chem
Mater 2000;12:1049–52.
[27] Available from: http://www.thomas-swan.co.uk/pages/nmframe-
set.html, http://www.nanocyl.com/.
[28] Available from: http://www.cus.cam.ac.uk/~cs266/list.html.
[29] Ajayan PM, Stephan O, Colliex C, Trauth D. Aligned carbon
nanotube arrays formed by cutting a polymer resin-nanotube
composite. Science 1994;265:1212–4.
[30] Ren Y, Li F, Cheng HM, Liao K. Tension–tension fatigue
behavior of unidirectional single-walled carbon nanotube rein-
forced epoxy composite. Carbon 2003;41:2177–9.
[31] Fiedler B, Gojny FH, Wichmann MHG, Nolte M, Schulte K.
Fundamental aspects of nano-reinforced composites. Special
Issue: 60th birthday Prof. K. Friedrich, Compos Sci Technol
2005 [accepted].
[32] Gojny FH, Wichmann MHG, Fiedler B, Kinloch I, Bauhofer W,
Windle AH, Schulte K. Evaluation and identification of electrical
and thermal conduction mechanisms in carbon nanotube/epoxy
composites 2005 [in press].
[33] Meguid SA, Sun Y. On the tensile and shear strength of nano-
reinforced composite interfaces. Mater Design 2004;25:289–96.
[34] Hsiao KT, Alms J, Advani SG. Use of epoxy/multi-walled carbon
nanotubes as adhesive to join graphite fibre reinforced polymer
composites. Nanotechnology 2003;14:791–3.
[35] Gojny FH, Wichmann MHG, Fiedler B, Schulte K. Influence of
nano-reinforcement on the mechanical and electrical properties of
conventional fibre reinforced composites. Composites A Applied
Science and Manufacturing 2005 [in press], Online available.